You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When we rank too many items over too many features, the redis read request becomes too large:
20:20:22.701 ERROR org.http4s.server.service-errors - Error servicing request: POST /rank from 127.0.0.1
redis.clients.jedis.exceptions.JedisDataException: ERR Protocol error: invalid bulk length
at redis.clients.jedis.Protocol.processError(Protocol.java:96)
at redis.clients.jedis.Protocol.process(Protocol.java:137)
at redis.clients.jedis.Protocol.read(Protocol.java:192)
at redis.clients.jedis.Connection.readProtocolWithCheckingBroken(Connection.java:316)
at redis.clients.jedis.Connection.getOne(Connection.java:298)
at redis.clients.jedis.Connection.executeCommand(Connection.java:123)
at redis.clients.jedis.Jedis.mget(Jedis.java:818)
at io.findify.featury.connector.redis.RedisStore.$anonfun$read$3(RedisStore.scala:47)
at apply @ io.findify.featury.connector.redis.RedisStore.$anonfun$read$2(RedisStore.scala:47)
at flatMap @ io.findify.featury.connector.redis.RedisStore.$anonfun$read$2(RedisStore.scala:47)
at make @ ai.metarank.mode.inference.FlinkMinicluster$.resource(FlinkMinicluster.scala:13)
at make @ ai.metarank.mode.inference.FlinkMinicluster$.resource(FlinkMinicluster.scala:13)
at use @ ai.metarank.mode.inference.Inference$.$anonfun$run$4(Inference.scala:33)
at flatMap @ ai.metarank.mode.inference.api.RankApi.$anonfun$rerank$2(RankApi.scala:34)
at apply @ ai.metarank.mode.inference.api.RankApi.rerank(RankApi.scala:33)
at flatMap @ ai.metarank.mode.inference.api.RankApi.rerank(RankApi.scala:33)
at flatMap @ ai.metarank.mode.inference.api.RankApi$$anonfun$1.$anonfun$applyOrElse$1(RankApi.scala:25)
at flatMap @ fs2.Compiler$Target.flatMap(Compiler.scala:162)
So we should split such requests into smaller chunks and send them in parallel instead.
The text was updated successfully, but these errors were encountered:
When we rank too many items over too many features, the redis read request becomes too large:
So we should split such requests into smaller chunks and send them in parallel instead.
The text was updated successfully, but these errors were encountered: